How dbdeux Works

dbdeux is a fully managed platform. You focus on writing dbt models. We handle everything else: execution, scaling, security, and monitoring.

High-Level Architecture

dbdeux is composed of three layers, each designed for reliability and performance:

1. Web Application

The browser-based interface where your team writes, tests, and deploys dbt models. It includes:

• The Cloud IDE with code editing, file management, and Git operations
• The DAG Explorer for visual lineage navigation
• Dashboards for monitoring runs, tests, and pipeline health
• Settings and administration for managing connections, teams, and environments

2. API Layer

The intelligent backend that orchestrates your entire dbt workflow:

• Manages projects, connections, and user permissions
• Queues and tracks dbt runs
• Processes schema snapshots and computes diffs
• Handles notifications and integrations
• Serves lineage data and search results

3. Compute Engine

The serverless execution layer that runs your dbt commands:

• Spins up isolated containers for each run (no cross-contamination between runs)
• Automatically selects the correct warehouse adapter
• Scales horizontally based on demand (10 concurrent runs or 1,000, same performance)
• Cleans up resources after each execution

How a Run Executes

When you click Run in the IDE or a scheduled job triggers:

1. Queue: The run request is queued with your project configuration
2. Provision: A fresh, isolated container spins up in seconds
3. Prepare: Your dbt project is loaded with the correct packages and adapter
4. Execute: The dbt command runs against your warehouse
5. Analyze: Results are parsed, schema changes detected, lineage updated
6. Report: Results appear in the UI, notifications fire, and the container is cleaned up

The entire process is fully managed. You never need to provision servers, manage containers, or worry about scaling.

Data Flow

Your data stays in your warehouse at all times. dbdeux never copies or stores your actual data:

Your Browser → dbdeux Platform → Your Warehouse
     ↑                                    |
     └────────── Results ─────────────────┘

• Queries execute in your warehouse using your credentials
• Only metadata flows through dbdeux (schema info, row counts, execution status)
• Your data never leaves your warehouse or transits through our servers

Multi-Tenant Isolation

Each organization gets complete isolation:

• Separate compute resources (no shared containers between organizations)
• Isolated data stores (your metadata is never co-mingled)
• Independent scaling (your workload does not affect others)
• Dedicated encryption keys per organization